A replicated discussion?

[Martin Taylor 931208 16:00]

Apropos the extended discussion on feedforward, predictive control,
adaptation ..., I have just managed to borrow the book recommended by Hans
Blom (Self-Tuning Systems: Control and Signal Processing, P.E. Wellstead
and M.B.Zarrop, Wiley, 1991). It uses the standard control engineer's
diagram of the loop, which we know how to map onto the PCT diagram, but
early in the introduction (P8) there appears the following quote,
which seems to describe the argument that has been going on in CSG-L:

"One of the traditional motivations for feedback...is that it is possible
to reduce the sensitivity of the overall system to parameter changes
in the feedback loop by using high feedback gains. It is therefore
reasonable to ask why we should use a potentially complex adjustment
mechanism to compensate for system change [i.e., environmental change.
MMT] when high gain feedback can desensitize the overall performance
to such changes. Original work on high gain feedback has led to the
area known as _robust control_. The proponents of robust control argue
strongly against the need for self-tuning and the debate on the valid
areas of application for self-tuning and robust control will undoubtedly
run and run. For the purpose of this text we assert that self-tuning is,
at least, a valid alternative to robust control, and, at best, it can
handle applications which are not tractable using robust control."

Sound familiar?

Martin